ExPart3D : Exclusive part aggregation for zero-shot point cloud learning

doi:10.1016/j.eswa.2026.132370

ExPart3D : Exclusive part aggregation for zero-shot point cloud learning

2026-04-07

0

PRE

AI

DOI：10.1016/j.eswa.2026.132370

Original

Original request for help

Save

Abstract

En 中文

• Novel framework discovers exclusive object parts for 3D point cloud ZSL. • Effective dual-branch aggregation: pooling-based and graph-based methods. • Achieves state-of-the-art on ModelNet40, McGill, and ScanObjectNN datasets.

Keywords:

ExPart3D

zero-shot learning

3D point cloud

object parts

dual-branch aggregation

ExPart3D : Exclusive part aggregation for zero-shot point cloud learning

Abstract

Journal

Organization